The lack of security in cordless telephones (the kind you might have in your home--not mobile cellular phones) is legendary. Perhaps you have even picked up a neighbor's conversation on your cordless phone and wondered whether your neighbor could eavesdrop on you. Imagine what would happen if you worked in a large organization--a hospital, for example--that had decided to use cordless phones extensively. The idea of using such phones instead of the beepers that physicians and other hospital personnel use so routinely seems attractive. But it quickly loses its appeal when you realize that, with so many cordless phones in a small area, others would be all too likely to listen in on your conversations.

That's where SpectraLink Corp comes in. It builds the Pocket Communications System (PCS) for facilities whose personnel need to stay in touch with one another, even though they are constantly on the go throughout a limited area. With PCS, a facility can enhance its private branch exchange (PBX) or Centrex system in a way that allows users to access all system features from pocket-sized cordless phones powered by rechargeable batteries. The technology that makes PCS possible is spread-spectrum communication. Thanks to spread spectrum, large numbers of PCS users can operate their cordless phones simultaneously with uncompromised security and without interfering with one another.

Spread spectrum is indeed a marvelous technology. Cloaked in secrecy for many years because of its origins in high-security military systems, it is finding new celebrity as the rising star of wireless communications.

SpectraLink's cordless phones do not require FCC licensing for two reasons: The transmitters use low power, and the system operates in a 902- to 928-MHz band. The FCC designates this band for unlicensed, low-power industrial, scientific, and medical (ISM) use. Transmissions in the ISM bands use spread-spectrum techniques, and, as a consequence, multiple signals can be present in the same frequency band at the same time without interfering with each other. Moreover, the system offers greater immunity to noise than do cordless phones that use narrowband technology.

Industry observers, even in the nontechnical press, have widely touted spread spectrum's signal/noise and data-security advantages. In fact, the enthusiasm of nontechnical journalists is running so high that it may be positioning the technique for a fall. Those who have embraced spread spectrum without becoming familiar with its exquisite complexity and arcane nuances should remember the old adage about claims that sound "too good to be true": Even if much of the hype surrounding spread spectrum isn't downright false, it is misleading.

Advantages not automatic

Yes, spread spectrum inherently provides high data security. But not all implementations take advantage of this feature. Yes, spread spectrum can provide excellent noise immunity. But blindly applying the technology fails to guarantee S/N ratios better than those of narrowband communication systems; incorrectly applied, spread spectrum can even cause inferior noise performance.

Compared with narrowband communications, spread spectrum promises vastly more efficient spectrum use. But phasing spread spectrum into the existing cellular network poses gargantuan problems. The equally unattractive alternatives are irreconcilable spectrum-allocation conflicts or the instant obsolescence of billions of dollars' worth of equipment in the hands of the public and cellular-service providers.

Spread spectrum is difficult to understand. That factor is one of the most potent militating against the tech- nology's rapid takeover of wireless voice and data communication. Deciding whether narrowband or spread-spectrum communication is best for an application requires a host of complex trade-offs. These trade-offs involve such factors as the nature of the information--for example, is it voice or data? In data communication, the delays introduced by error-correcting protocols are usually not serious. However, similar delays become unacceptable in two-way voice communication--even when the voice is encoded in digital form.

The nature of the interfering signals is another factor that affects spread-spectrum's suitability and determines whether an application should use direct sequence or frequency hopping (spread spectrum's two main forms). Although spread-spectrum communications might seem to be immune to narrowband interference, such interference can be devastating. A frequency-hopping system that hops to a frequency occupied by a narrowband signal can lose all data until its next hop. But if the hops occur often enough and are of short-enough duration, the loss of information may be acceptable.

Analyzing an application often involves not only solving complicated math, but also finding guidance. So-called spread-spectrum experts often disagree. In some quarters, the debate over the superiority of direct sequence vs frequency hopping has assumed the dimensions of a religious war. Dispassionate experts--when you can find them--often refuse to make blanket recommendations about technical choices. Moreover, don't be surprised if some experts won't even spell out the factors you should consider in reaching a decision. The most-uttered phrase in discussions of spread-spectrum communications is, "It depends."

Spread spectrum owes its surge in popularity not only to its unique attributes but also to modern semiconductor technology. Even five years ago, implementing a spread-spectrum receiver for a consumer application was not economically feasible. But higher levels of integration have driven down the size and cost of the hardware to the point where handheld spread-spectrum cellular phones are both technically and economically practical.

Of the two basic spread-spectrum technologies--frequency hopping and direct sequence--frequency hopping is the easier to understand. In frequency hopping, the information is modulated onto a carrier derived from a frequency synthesizer. A pseudorandom sequence (PRS) determines the synthesizer's output frequency.

The carrier is restricted to a finite set of predetermined frequencies, all of which must be used for roughly equal portions of the time. In the 902- to 928-MHz ISM band, a frequency-hopping transmitter must use at least 50 frequencies. In the higher frequency ISM bands, a transmitter must use at least 75 frequencies. In the lower ISM band, a transmitter must occupy a single frequency for no more than 0.4 sec in a 20-sec interval. In the higher bands, a transmitter must occupy a single frequency for no more than 0.4 sec in a 30-sec interval.

Nothing in the FCC's rules dictates that a transmitter shall dwell continuously at a single frequency for as long as 0.4 sec, however. Indeed, the 0.4 sec that a transmitter spends at a frequency over the course of 20 or 30 sec usually consists of many hops to that frequency. In any system, the duration of all hops is constant.

The receiver for the frequency-hopping transmissions must be synchronized with the transmitter. Like the transmitter frequency, the receiver frequency is under the control of a synthesizer. The same PRS drives both synthesizers.

The PRS is at the heart of the data security inherent in frequency hopping. Because a sequence can allow multiple hops to each frequency, a system that uses 50 frequencies can hop many more than 50 times during a single repetition of a sequence. Because of the length of the sequence and the large number of possible next frequencies at each hop, the number of possible sequences is enormous. Thus, it is unlikely that, without "inside information," someone could set up a receiver to track a frequency-hopping transmitter.

The system's noise advantages are the result of its use of a frequency band much wider than that required to transmit the information. Narrowband noise may corrupt the signal for short periods, but in voice transmissions, those periods tend to be so short that they do not affect intelligibility. Similarly, in data transmission, although occasional packets may be corrupted, error-correction schemes generally allow recovery of the data without significantly reducing the data rate.

You might think that using a wide frequency band wastes spectrum space. It doesn't, however, because several frequency-hopping systems can occupy the same frequency band simultaneously without significantly interfering with each other. In this situation, although multiple transmitters sometimes hop to the same frequency at the same time and, therefore, briefly interfere, the interference is slight if the number of transmitters is not excessive.

Simple, elegant, unfathomable

The direct-sequence approach is at once simpler, more elegant, and harder to understand than frequency hopping. Even though the system is simpler and more elegant, it isn't necessarily better or worse. Like frequency hopping, direct sequence uses a PRS. Unlike frequency hopping, it does not use frequency synthesizers. In contrast with frequency hopping, direct-sequence combines the (binary) data with a pseudorandom binary signal in an exclusive-OR gate ahead of the modulator. The highest frequency present in the PRS (the so-called chipping rate) is at least as high as the highest frequency in the data and is usually considerably higher.

The result of combining the data with the PRS is that the bandwidth of the signal that modulates the carrier is much greater than it otherwise would be; the modulated carrier covers a much greater range of frequencies; and less energy is present at any frequency than would be present without use of the PRS. The total energy contained in the modulated carrier is unaffected by combining the data with the PRS, however.

To recover the original signal, the down-converted receiver output passes through a correlator. The same PRS that was X-ORed with the original data also feeds into this correlator. Other direct-sequence signals are uncorrelated with this PRS, making it possible to recover the original data. Indeed, just as with frequency hopping, multiple direct-sequence signals can occupy the same frequency band simultaneously without interfering with each other. Moreover, the system rejects noise because the correlator behaves, in a sense, like a sharply tuned filter that rejects noise uncorrelated with the PRS. Neither narrowband nor truly random noise correlates with a PRS.

Although the use of frequency synthesizers might make frequency-hopping systems more expensive than direct-sequence systems, general agreement on this point is lacking. Furthermore, frequency-hopping transmitters emit signals that, at any instant, cover only a narrow frequency range. As a result, frequency-hopping systems might be able to use lower transmitter power to achieve S/N ratios equivalent to those of direct-sequence systems. If frequency hopping can provide S/N ratios equivalent to or better than those of direct sequence, frequency hopping could become the technology of choice in battery-powered applications. If you try to determine what direction the technology actually will take, however, the answer you get depends on whom you ask.

You can reach Senior Technical Editor Dan Strassberg at (617) 558-4205; fax (617) 558-4470.EDN BBS: EDNStras. Internet: ednstrassberg@mcimail.com

1.Gallant, John, "Digital Wireless Networks," EDN, March 4, 1993, pg 78.